DeepScience
All digests
ResearchersENNuclear Fusiondaily

[Nuclear Fusion] Daily digest — 287 papers, 0 strong connections (2026-05-04)

DeepScience — Nuclear Fusion
DeepScience
Nuclear Fusion · Daily Digest
May 04, 2026
287
Papers
10/10
Roadblocks Active
5
Connections
⚡ Signal of the Day
• ELM control dominates today's fusion papers, with three independent experimental and deployment studies each demonstrating a distinct method to suppress or pace edge instabilities that threaten plasma-facing components.
• Nitrogen seeding on EAST achieves full ELM suppression while boosting confinement by ~30% (H98 from 0.9 to 1.2), vertical jogging on DIII-D cuts per-event energy loss from ~10% to under 1%, and an FPGA-deployed neural network enables sub-millisecond predictive ELM forecasting—all in one day's data.
• Watch for whether nitrogen seeding's confinement benefit survives longer pulses and high-Z wall conditions on other machines; the mechanism (drift-type edge mode replacing peeling-ballooning instability) needs multi-machine validation before ITER extrapolation.
📄 Top 10 Papers
Characterization of ELM Pacing via Vertical Jogs on DIII-D
Displacing the plasma vertically at 20 Hz forces ELMs to fire four times more often but with far less energy each time—dropping per-event stored-energy loss from roughly 10% to under 1% and cutting peak divertor heat flux by about half. The effect works because each artificial trigger fires before enough edge pressure can accumulate for a large burst. For ITER and future reactors, this technique requires only fast coil control with no impurity injection, making it a low-infrastructure route to keeping divertor heat loads within material limits.
██████████ 1.0 elm-control Preprint
Nitrogen-induced ELM suppression and confinement improvement in the EAST tokamak with a full metal wall
Injecting nitrogen gas into the EAST tokamak simultaneously eliminates large ELM bursts and raises the global energy confinement factor (H98) from ~0.9 to ~1.2, demonstrating that high performance and wall protection need not be traded off against each other. Gyrokinetic simulations suggest nitrogen drives a drift-type turbulence mode at the plasma edge (20–50 kHz) that stabilizes the pressure-driven instability responsible for ELM crashes. The result is notable for being achieved on a full-metal-wall device, the configuration planned for ITER, though reproducibility across machines and over longer pulse lengths remains to be established.
██████████ 0.9 elm-control Preprint
FPGA-Accelerated Real-Time Diagnostics at DIII-D Using the SLAC Neural Network Library for ML Inference
A neural network running on dedicated FPGA hardware inside the DIII-D plasma control system can forecast ELMs in real time from 160 turbulence measurement channels, with latency low enough to trigger preventive coil responses before the instability fires. A key practical feature is that network weights can be updated on the fly without rebuilding the hardware design, allowing the model to adapt as plasma conditions change. Real-time predictive ELM suppression is essential for future reactors where a single large unmitigated ELM could erode tungsten divertor tiles beyond safe limits.
██████████ 0.9 elm-control Preprint
The L-H transition in tokamaks: power threshold, density minimum and toroidal-field asymmetry
Three-dimensional fluid simulations using the GBS code reproduce from first principles why tokamak plasmas enter high-confinement mode at lower heating power when the magnetic field points in one direction versus the other—a long-standing experimental observation without a clear mechanism. The simulations show that turbulence spontaneously generates a sheared E×B flow layer that suppresses heat transport, with the directional asymmetry caused by finite particle collision rates that break time-reversal symmetry. A quantitative, physics-based L-H threshold prediction matters because overestimating the required heating power in ITER could mean missing high-confinement operation entirely.
██████████ 0.9 turbulence-modeling Preprint
Quasilinear flux model consistent with gyrokinetic ordering
A new analytical formula for turbulent heat and particle transport in tokamaks derives the amplitude of turbulent fluctuations directly from fundamental gyrokinetic ordering relations, removing the need for expensive nonlinear simulations to set a free calibration parameter. The resulting model correctly reproduces both the wavenumber dependence and absolute magnitude of ion heat transport when benchmarked against existing nonlinear simulations for standard test-case parameters. Faster and self-consistent transport models are critical for predicting plasma performance in ITER and for rapid scenario optimization in future devices.
██████████ 0.9 turbulence-modeling Preprint
Electrothermal Dynamics of Cold Front in Impure Tokamak Plasmas
A reaction-diffusion model shows that when a cold temperature front sweeps through tokamak plasma during impurity-driven radiative collapse, it produces sharp spikes in local current density—a previously underappreciated route to disruption. The spikes arise because the curvature of the resistivity profile within the thin cold layer acts as a strong source term in the current diffusion equation, creating current over-density just ahead of the front and depletion behind it. This finding suggests that impurity-driven cold fronts can independently seed disruption-causing MHD instabilities even before a full thermal quench, pointing to a new target for early disruption detection.
██████████ 0.8 plasma-disruption Preprint
A simple model of current ramp down in the ITER tokamak
A 1D cylindrical MHD model shows that ITER's planned 60-second plasma current ramp-down is stable against disruptions provided the plasma is sufficiently hot at the start of the ramp; the only dangerous instability is the m=2/n=1 classical tearing mode. Ramp-down timescales significantly shorter than 60 seconds excite this mode, which can lock to the vacuum vessel and trigger a full disruption. The result provides a concrete lower bound on ramp-down speed for ITER operations and a framework for testing whether auxiliary heating during the ramp can buy additional margin.
██████████ 0.8 plasma-disruption Preprint
Non-thermal electron cyclotron emission during runaway plateau in tokamak disruptions from an analytic hot plasma dispersion tensor
Closed-form analytic expressions are derived for the electromagnetic emission and absorption coefficients of runaway electron beams using a Gaussian pitch-angle distribution, replacing previously numerical-only calculations. The analytic dispersion tensor allows rapid characterization of runaway electron energy and pitch angle from measured radiation spectra, and it has been verified against two independent numerical codes (KIAT and SYNO). Fast, analytic runaway diagnostics matter because delayed identification of runaway plateau properties limits the time available to deploy mitigation—such as shattered pellet injection—before the beam strikes and damages plasma-facing components.
██████████ 0.8 plasma-disruption Preprint
Improved n=1 Empirical Error Field Penetration Threshold Scaling with Ohmic and L-Mode Conventional Tokamak Plasma Discharges
An updated empirical scaling law predicts at what level of magnetic field imperfection (error fields) a tokamak plasma will lock its rotation and risk disruption, now fitted to an expanded database including J-TEXT and new JET data across eight machines. The analysis is deliberately restricted to Ohmic and L-mode plasmas—the regime most vulnerable to error field penetration—improving the relevance of projections to ITER startup scenarios. Error field correction coils must be energized correctly from ITER's first plasma, making an accurate threshold prediction a practical Day-1 operations requirement.
██████████ 0.7 plasma-disruption Preprint
Energy-differential measurement of the $^{\mathrm{nat}}$C(n,p) and $^{\mathrm{nat}}$C(n,d) reactions at the n_TOF facility at CERN
Measurements at CERN's neutron time-of-flight facility show that the standard nuclear data libraries used worldwide (ENDF, JEFF, JENDL) significantly mis-predict how 14 MeV fusion neutrons interact with carbon, producing protons and deuterons at rates that differ substantially from library values. Carbon is a pervasive impurity and historical wall material in fusion devices, and accurate cross-sections at fusion neutron energies are needed to calculate activation inventories, helium bubble production that causes material embrittlement, and tritium generation in blankets. The discrepancies found indicate that current activation codes using these libraries may be systematically wrong for carbon-containing first-wall and blanket components.
██████████ 0.6 first-wall-materials Preprint
🔬 Roadblock Activity
Roadblock Papers Status Signal
Plasma Turbulence Modeling 27 Active Two complementary advances today: first-principles fluid simulations reproduce the L-H transition power threshold asymmetry without free parameters, and a new quasilinear transport model eliminates the need for nonlinear calibration runs by deriving saturation amplitude from gyrokinetic ordering.
Engineering Gain (Q > 1) 17 Active Activity is broad but diffuse; the strongest fusion-relevant signal is FPGA-accelerated real-time ELM forecasting, which reduces auxiliary system dead time and could improve effective plant utilization factor.
Plasma-Wall Interaction 12 Active Carbon neutron cross-section discrepancies with major nuclear libraries were identified at fusion-relevant energies, flagging potential errors in first-wall activation and helium embrittlement calculations for carbon-containing components.
Plasma Disruption Prevention 12 Active Three mechanistic papers advance disruption physics simultaneously: cold-front-induced current spikes as a novel disruption seed, analytic runaway electron emission diagnostics for faster identification, and an ITER current ramp-down stability bound.
Long-Pulse Confinement 11 Active No dedicated long-pulse result today, but the nitrogen ELM suppression study on EAST—demonstrating simultaneous confinement improvement and ELM-free operation on a metal wall—is the most relevant signal for sustained high-performance operation.
ELM Control 6 Open A notably strong day: nitrogen seeding, vertical jogging, and FPGA-accelerated ML forecasting each address ELM suppression or pacing from distinct physical and engineering angles, with all three yielding positive experimental or deployment results.
Divertor Thermal Management 4 Open Vertical jog ELM pacing reducing peak divertor heat flux by roughly half on DIII-D is the main signal, with nitrogen seeding on EAST providing a secondary route via complete ELM elimination.
High-Temperature Superconducting Magnets 2 Low Light activity today; the stellarator coil non-planarity study has marginal relevance to HTS magnet design constraints but no direct magnet performance or quench physics results appeared.
First Wall Materials 2 Low New n_TOF measurements reveal significant discrepancies with ENDF/JEFF/JENDL libraries for carbon neutron reactions at fusion-relevant energies, indicating a gap in the nuclear data underpinning first-wall activation and damage calculations.
Tritium Breeding 1 Low Minimal direct activity today; only an indirect connection through the carbon neutron cross-section measurements, which affect activation calculations in carbon-bearing blanket materials.
View Full Analysis
DeepScience — Cross-domain scientific intelligence
Sources: arXiv · OpenAlex · Unpaywall
deepsci.io